Are Young and Old Workers Harmful for Firm Productivity? - SSRN

4 downloads 109 Views 430KB Size Report
might be more reluctant to invest in training for older workers because they .... construction (F), wholesale and retail trade, repair of motor vehicles, ..... and repair of motor vehicles and motorcycles ; retail sale of automotive fuel (NACE 50) ;.
DISCUSSION PAPER SERIES

IZA DP No. 3938

Are Young and Old Workers Harmful for Firm Productivity? Thierry Lallemand François Rycx

January 2009

Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor

Are Young and Old Workers Harmful for Firm Productivity? Thierry Lallemand Université Libre de Bruxelles, DULBEA

François Rycx Université Libre de Bruxelles, CEB, DULBEA and IZA

Discussion Paper No. 3938 January 2009

IZA P.O. Box 7240 53072 Bonn Germany Phone: +49-228-3894-0 Fax: +49-228-3894-180 E-mail: [email protected]

Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post World Net. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.

IZA Discussion Paper No. 3938 January 2009

ABSTRACT Are Young and Old Workers Harmful for Firm Productivity?* This paper investigates the effects of the workforce age structure on the productivity of large Belgian firms. More precisely, it examines different scenarios of changes in the proportion of young (16-29 years), middle-aged (30-49 years) and old (more than 49 years) workers and their expected effects on firm productivity. Using detailed matched employer-employee data, we find that a higher share of young (old) workers within firms is favourable (harmful) for firm value added per capita. Results also show that age structure effects on productivity are stronger in ICT than in non-ICT firms.

JEL Classification: Keywords:

J21, J31, L25

firm performance, workforce age structure, demographic changes

Corresponding author: François Rycx Université Libre de Bruxelles CP 140, Av. F.D. Roosevelt 50 B-1050 Brussels Belgium E-mail: [email protected]

*

This paper is produced as part of a research contract for the European Commission (DG Employment, No. VT/2005/92) entitled “Study and Conference on European Labour Market Analysis using Firm-Level Panel Data and Linked Employer-Employee Data” (http://cep.lse.ac.uk/leed/). Financial support from the National Bank of Belgium is gratefully acknowledged. We also would like to thank Statistics Belgium for giving access to the Structure of Earnings Survey and the Structure of Business Survey. The usual disclaimer applies.

1. INTRODUCTION

There is strong evidence that the average age of the population is steadily increasing in most European countries. This ongoing demographic change might have serious consequences for the management of human resources within firms. This paper aims to examine one of these effects. More precisely, we want to investigate whether the age structure of the workforce has an impact on the productivity of Belgian firms. This issue is of major importance for economic policy as most production in Belgium draws on prime-aged workers (25-54 age group). Indeed, Belgium’s employment rate for people aged 55-64 is one of the lowest in the OECD area (34% in 2007 (OECD, 2008)). The withdrawal of older people from the Belgian labour market is mainly due to the existence of attractive (to both employers and employees) early retirement schemes. It also derives from the fact that older workers are often considered by employers as more costly and less productive than prime-aged workers. Moreover, Belgium is characterized by a low employment rate and a high unemployment rate (19% in 2007 (OECD, 2008)) for people aged 15-24. Most young unemployed have a low level of education (OECD, 2007). This is not surprising as firms often complain that young workers have a lack of relevant skills and therefore prefer to opt for middle-aged individuals. Finally, let us notice that the rapid ageing process of the population requires policy-makers in most European countries, and certainly in Belgium, to find solutions to reduce the pressure that the demographic dependency ratios put on public finances. One way to deal with this issue is to increase the employment rate of people aged 55-64 (the European target stands at 50% for 2010). Another (complementary) solution would be to improve the employment prospects of young people. These changes in the workforce age structure may have important consequences for the productivity of firms. Yet, the empirical evidence on this issue is surprisingly limited. Moreover, as far as we know, this question has never been investigated for Belgium. 2

The theoretical literature on the relation between age and productivity is ambiguous. On the one hand, several theories suggest that older workers are more productive. Mincer (1974), for instance, argues that older workers have more job experience and know-how, which increases their individual performance. There is also a higher probability that they have been assigned to their best position in the firm organization (Jovanovic, 1979). Moreover, they are more likely to have correctly matched their job preferences with the employer’s requirements (Johnson, 1978). Autor et al. (2003) expect a rise of the performance of older workers over time. The point is that during the last decade the demand for interactive skills on the US labour market (i.e. abilities which do not generally vary with age) has increased more than the demand for problem-solving and mathematical abilities (i.e. skills that are supposed to be declining with age). On the other hand, there are multiple factors suggesting that younger workers are more productive. One of the most frequently cited arguments is that a worker’s health tends to deteriorate over the life cycle (e.g. diseases, absenteeism, body strength, depression, etc.). In addition, it is argued that cognitive abilities generally decrease with age.1 This may result in a lower productivity level of older workers, unless their job experience and specific knowledge compensate for their inferior cognitive skills.2 Besides, young people are thought to be more motivated to exert higher effort since they want to give a good signal to their employer (Grund and Westergaard-Nielsen, 2005). In contrast, older people might be less willing to invest in training programmes since they are closer to retirement and cannot learn new skills well (Hayward et al., 1997). This is in general the perception that employers have of older workers (Itzin et al., 1994). Moreover, employers might be more reluctant to invest in training for older workers because they have a shorter period of time to benefit from on-the-job training (Brooke, 2003; Prskawetz et al., 2006).

1

Cognitive abilities reflect inter alia numerical capabilities and verbal, reasoning and problem solving abilities.

2

See Skirbekk (2003) for a broad literature review on this issue.

3

Nevertheless, the reverse argument might also hold since young workers change jobs more often, which might reduce the return of employers to on-the-job training (Taylor and Urwin, 2001). The empirical literature regarding the impact of the workforce age structure on the productivity of firms is still limited. Most papers, relying on matched employer-employee data, report a positive and hump-shaped relationship between workforce age and firm performance. In general, strong decreases in productivity are observed after the age of 50.3 However, two studies found that productivity peaks at 55 years or more (Hellerstein and Neumark, 1995; Hellerstein et al., 1999). Yet, these results may be attributed to the poor quality of data in the former study and to the use of output as an estimate of productivity in the latter. In general, studies investigating the age-productivity relationship use workers’ mean age or shares of workers by age groups as an indicator of the age structure of the workforce, while the most common measure of firm productivity is firm’s value added. Yet, many papers rely on cross-sectional data and/or focus on a particular sector of economic activity (in general, the manufacturing and mining sector). This paper is the first to investigate the effects of the workforce age structure on firm productivity in the Belgian private sector. Our main objective is to show whether shifts in the current workforce age structure would be beneficial or detrimental for the productivity of Belgian firms. More precisely, we examine different scenarios of changes in the proportion of young and old workers and their effects on firm productivity. This paper contributes significantly to the existing literature as it is one of few to: i) focus on more than one or two industries, ii) show the impact on productivity of all possible changes in the proportion of 3

See Andersson et al. (2002) for Sweden, Aubert and Crépon (2003) for France, Dostie (2006) for Canada,

Grund and Westergaard-Nielsen (2005) for Denmark, Haegeland and Klette (1999) for Norway, Haltiwanger et al. (1999) for US, Hellerstein and Neumark (2004) for US, Malmberg et al. (2005) for Sweden, Prskawetz et al. (2006) for Austria.

4

young, prime-aged and old workers at the firm level, iii) distinguish between ICT and nonICT firms, and iv) deal with the potential endogeneity problem of the workforce age structure.4 In the first part of the paper, we use two detailed matched employer-employee data sets for the years 1995 and 2003. It enables us to test the stability of the results over time and to take into account (to a certain extent) the potential cohort effects that could drive age effects. Our data sets derive from the combination of the Structure of Business Survey (SBS) and the Structure of Earnings Survey (SES). The former provides firm-level information on financial variables and in particular on the productivity of the workforce (e.g. value added). The latter contains detailed information on individual workers (e.g. gross hourly wages, age, education, sex, and occupation) and on firm characteristics (e.g. sector of activity, level of wage bargaining, and firm size). In the second part of the paper, we perform sensitivity and robustness tests. First, we examine whether results may differ across firms that use ICT intensively and those that do not. Next, we address two potential problems linked to the robustness of OLS estimates of age effects. To do so, we rely on a unique small panel data set of firms built from a combination of the SES and SBS in 1995 and 2003. On the one hand, we try to correct for the endogeneity of workforce age structure using its lagged values. On the other hand, we try to control for firm fixed effects by estimating our model in first differences. The plan of the rest of the paper is as follows. We first present the data set in section II and describe our methodology in the next section. In section IV, we explain our sampling 4

Recently, several papers have tried to overcome this problem. For papers using IV estimations to correct for the

bias of simultaneity, see: Andersson et al. (2002), Aubert and Crépon (2003) and Malmberg et al. (2005). In contrast, Hellerstein and Neumark (2004) and Dostie (2006) follow the methodology developed by Olley and Pakes (1996) and Levinshon and Petrin (2003), which consists in proxying firm’s unobservable productivity shocks by either investments or intermediate inputs.

5

scheme and show several descriptive statistics. Section V is devoted to the presentation and discussion of the main results. In sections VI and VII, we run sensitivity and robustness tests. We draw some conclusions in the last section.

2. DATA SET

Each final data set (for 1995 and 2003) is based upon a unique combination of two large-scale data sets. The first, conducted by Statistics Belgium5, is the Structure of Earnings Survey (SES). It covers all Belgian firms employing at least 10 workers and with economic activities within sections C to K of the NACE Rev.1 nomenclature. It thus encompasses the following sectors: mining and quarrying (C), manufacturing (D), electricity and water supply (E), construction (F), wholesale and retail trade, repair of motor vehicles, motorcycles and personal and household goods (G), hotels and restaurants (H), transport, storage and communication (I), financial intermediation (J), and real estate, renting and business activities (K). The survey contains a wealth of information, provided by the management of the firms, both on the characteristics of the firms (e.g. sector of activity, number of workers, level of collective wage bargaining, type of economic and financial control, region) and on the individual employees (e.g. age, educational level, tenure, gross earnings, paid hours, sex, occupation, type of contract). The SES provides no financial information. This is why the SES has been combined with the Structure of Business Survey (SBS). It is a firm-level survey, conducted by Statistics Belgium, with a different coverage than the SES in that it includes neither the financial sector

5

According to the instructions given by Eurostat (E-U regulation Nr. 2744/95).

6

(NACE J)6 nor firms with less than 20 employees. Both data sets have been merged by Statistics Belgium using firms’ social security number. The SBS provides firm-level information on financial variables such as sales, value added, value of production and gross operating surplus.

3. METHODOLOGY

In order to investigate the effects of the workforce age structure on firm productivity, we estimate a regression where the dependent variable is the productivity per worker. The age structure of the workforce within a plant is measured by several age shares variables (3 categories). In addition, we control for a large list of variables influencing firm performance.

ln Pj = α + β 1 ln A1 j + β 2 ln A2 j + β 3 ln A3 j + γ X j + δ Y j+ε j

(1)

Pj is the productivity of firm j and is measured by the value added per employee. The value

added (at factor costs) per employee is obtained by dividing the firm’s annual gross operating income (plus subsidies, minus indirect taxes) by the number of workers in the firm. We split employees of a firm into three age groups: 50%) 1.90 Privately owned firm 91.61 Other 6.40 Sector Mining and quarrying (C) 0.44 Manufacturing (D) 52.43 Electricity, gas and water supply (E) 0.86 Construction (F) 3.75 Wholesale and retail trade; repair of motor vehicles (G) 21.95 Hotels and restaurants (H) 1.23 Transport, storage and communication (I) 5.86 Financial activities (J) 0 Real estate, renting and business activities (K) 13.48 Average number of observations per firm 34.73 Number of firms 691 + The descriptive statistics refer to the weighted sample. 1 Individual gross hourly wages include overtime paid, premiums for shift work, work. 2 CA stands for collective labour agreement.

2003 SD 42.20 14.3 13.4 8.9 3.63 1.54 3.68 3.18 4.60 25.10 1.82 0.81 34.44 24.12 594.53

Mean 69.33 21.94 61.42 16.64 38.49 9.54 14.64 4.46 9.67 29.75 11.38 1.88 51.88 13.46 266.49

SD 49.33 12.92 12.87 10.51 3.53 1.51 3.81 3.12 4.89 29.84 1.93 0.82 35.76 17.96 363.46

15.95 20.55 63.50 63.77 36.23 0 0.15 0.98 96.45 2.42 0.37 46.66 0 8.61 16.64 2.13 8.46 0.97 16.17 25.05 1,204 night work and/or weekend

24

Table 2: OLS estimates of age structure effects on firm productivity 1995 (1) (2) (3) Log share